introduction to optical communications copyrighted · pdf fileintroduction to optical ... the...

1Introduction to OpticalCommunications

1.1 Evolution of Lightwave Technology

The invention of the solid state ruby laser (acronym for light amplification by stimulatedemission of radiation) in May 1960 [1] and the He-Ne gas laser [2] in December 1960has led to some wide-ranging and very significant scientific and technological progress.This so-called ‘discovery of the century’, followed by the first use of semiconductorlasers [3–5] in communications, heralded the start of optical communications. The laserprovided a powerful coherent light source together with the possibility of modulation athigh frequency, and this opened up a new portion of the electromagnetic spectrum withfrequencies many times higher than those commonly available in radio communicationsystems. In addition the narrow beam divergence of the laser made enhanced free-spaceoptical transmission a practical possibility.

Since optical frequencies are of the order of 100 THz, and information capacityincreases directly with frequency bandwidth, the laser potentially offers a few order ofmagnitude increase in available bandwidth compared with microwave systems. Thus, byusing only a small portion of the available frequency spectrum, a single laser could, inprinciple, carry millions of telephone conversations or TV channels.

With the potential of such wideband transmission capabilities in mind, a number ofexperiments [6] using atmospheric optical channels were carried out in the early 1960s.These experiments showed the feasibility of modulating a coherent optical carrier waveat very high frequencies. However, the high cost of development for all the necessarycomponents, and the limitations imposed on the atmospheric channel by rain, fog, snowand dust make such high-speed systems economically unattractive. However, numerousdevelopments of free-space optical channel systems operating at baseband frequencieswere in progress for earth-to-space communications [7, 8].

It soon became apparent that some form of optical waveguide was required. By1963, bundles of several hundred glass fibres were already being used for small-scale

Optical CDMA Networks: Principles, Analysis and Applications, First Edition. Hooshang Ghafouri-Shiraz andM. Massoud Karbassian.© 2012 John Wiley & Sons, Ltd. Published 2012 by John Wiley & Sons, Ltd.

COPYRIG

HTED M

ATERIAL

2 Optical CDMA Networks

illumination, but these early fibres had very high attenuations and so their use as atransmission medium for optical communications was not considerable. Optical fibrescan provide a much more reliable and versatile optical channel than the atmosphere.Initially, the extremely large losses of more than 1000 dB/km observed in even thebest optical fibres made them appear impractical. In fact, to compete with existingcoaxial cable transmission lines, the glass fibre attenuation had to be reduced to lessthan 20 dB/km. It was in 1966 that C.K. Kao (2009 Nobel Laureate) and G.A. Hockman[9] speculated that these high losses were as a result of impurities in the fibre material,and that the losses could be reduced to the point where optical waveguides would be aviable transmission medium. This was realized in 1970 when Kapron, Keck and Maurer[10] of the Corning Glass Works fabricated a fibre having 20 dB/km attenuation. Atthis attenuation, repeater spacing for optical fibre links become comparable to those ofcopper systems, thereby making lightwave technology an engineering reality. A wholenew era of optical fibre communications was thus launched.

The ensuing development of optical fibre transmission systems grew from the com-bination of semiconductor technology, which provided the necessary light sources andphotodetectors, and optical waveguide technology upon which the optical fibre is based.The result was a transmission link that had certain inherent advantages over conventionalcopper systems in telecommunications applications. For example, optical fibres have lowertransmission losses and wider bandwidths as compared to copper wires.

This means that, with optical fibre cable systems, more data can be sent through onefibre, over longer distances, thereby decreasing the number of channels and reducing thenumber of repeaters needed over these distances. In addition, the low weight and thesmall hair-sized dimensions of fibres offer a distinct advantage over heavy, bulky wirecables in crowded underground city ducts. The low weight and small size are also ofimportance in aircraft where small lightweight cables are advantageous, and in tacticalmilitary applications where large amounts of cable must be unreeled and retrieved rapidly.

An especially important feature of optical fibres relates to their dielectric nature.This provides optical waveguides with immunity to electromagnetic interference, suchas inductive pick-up from signal-carrying wires and from lightning, and freedom fromelectromagnetic pulse effects, the latter being of particular interest for military applica-tions. Furthermore, ground loops are no longer an issue, fibre-to-fibre crosstalk is verylow and a high degree of data security is afforded since the optical signal is well confinedwithin the waveguide. Of additional importance is the advantage that silica is the principalmaterial of which optical fibres are made. This material is abundant and inexpensive sincethe main source of silica is sand.

The recognition of optical fibre advantages in the early 1970s created a flurry of activityin all areas related to optical fibre transmission systems. This development led to the firstlaboratory demonstration of optical communication with glass fibre in the early 1970s.Such a progress resulted in significant technological advances in optical sources, fibres,photodetectors and fibre cable connectors. Since then, research on material for opticalfibre transmission has made dramatic progress. Fibre loss was reduced from 1000 dB/kmto 20 dB/km in 1970 and to 4 dB/km in 1973. By using longer-wavelength transmissionthe optical fibre losses were further reduced to 2 dB/km in 1974, 0.5 dB/km in 1976 and0.2 dB/km by 1979.

Introduction to Optical Communications 3

8000

1

2

Atte

nuat

ion

(dB

/km

)

3

4

5

6

900 1000 1100 1200 1300

Wavelength (nm)

1400 1500 1600 1700 1800

Figure 1.1 Optical signal attenuation against wavelength

Also, a study of the spectral response of glass fibres showed the presence of low-loss transmissions at 850 nm, 1300 nm and 1550 nm as shown in Figure 1.1. Althoughthe early optical links used the 850 nm window, the longer wavelength windows exhibitlower losses, typically 0.5 dB/km at 1300 nm and 0.22 dB/km at 1550 nm. As a result,most modern links use 1300–1550 nm wavelength light sources.

New types of fibre materials have also been investigated [11–14] for use in the3 μm–5 μm wavelength bands. It was found that fluoride glasses have extremely lowtransmission losses at mid-infrared wavelengths (i.e. 0.2 μm < λ < 8 μm) with the lowestloss being around 2.25 μm. The material that has been concentrated on is a heavy-metalfluoride glass which uses ZrF4 as the major component. Although this glass potentiallyoffers intrinsic minimum losses of 1.01–0.001 dB/km, fabricating long lengths of thesefibres is difficult. Firstly, ultra-pure materials must be used to reach this low level. Sec-ondly, fluoride glass is prone to devitrification. Fibre-making techniques have to take thisinto account to avoid the formation of microcrystallites, which have a drastic effect onscattering losses.

1.2 Laser Technologies

The prospects for lower fibre losses at longer wavelengths led to intensive research onlasers and photodetectors. The advent of the semiconductor laser in 1962 meant that afast light source was available. The material used was gallium arsenide (GaAs) whichemits light at a wavelength of 870 nm. With the discovery of the 850 nm window, thewavelength of emission was reduced by doping the GaAs with aluminium (Al). Latermodifications included different laser structures to increase device efficiency and lifetime.As the longer wavelength windows exhibit lower losses, various materials – in particularInGaAsP/InP – were also investigated to produce devices for operation at 1300 nm and1550 nm. These efforts have also been successful. Commercial systems that used 1300and 1550 nm technologies appeared in early and mid 1980s, respectively. Semiconductor


sources are now available which emit at any one of the above wavelengths, with modu-lation speeds of several Gbits/s being routinely achieved.

The field of semiconductor lasers which has already reached a considerable level ofdevelopment has recently undergone more changes. Initially, semiconductor laser wave-lengths were at the infrared end of the spectrum. However, by shortening the wavelengthsit becomes possible to concentrate the light in a smaller area thus increasing the energydensity and producing a light source that is suited to optical data processing devices suchas optical disk and optical printers. As a result, research was carried out into reducing thewavelength of the semiconductor laser to within the visible spectrum. At present, laserdiodes emitting at 780 nm are available for use as light sources in digital audio disksand related devices. In addition, high-power semiconductor lasers (higher than 40 mW)emitting at 980 nm and 1480 nm are used as pump sources for erbium-doped fibre ampli-fiers [15]. At longer wavelengths (i.e. 1000–1600 nm) both bulk Fabry–Perot (FP) anddistributed feedback (DFB) lasers [16] and more recently multiple quantum well lasershave been fabricated and used successfully in long-haul fibre communication systems.Currently new optical sources, termed strained bulk and quantum well lasers, which offerbetter performance have been investigated intensively [17].

According to the results of a recent survey, laser techniques are used in a wide rangeof industrial fields such as measuring technologies, communications and data processing.Among laser light sources in general, there has been a notable increase in the use ofsemiconductor and carbonic acid gas lasers. New light sources that are currently at theresearch and development stage include the excimer laser, synchrotron-generated light,the free electron laser and strained bulk and quantum well lasers. Currently the mostpromising technological development in the field of optoelectronics is the excimer laser.This laser produces wavelength at the ultraviolet end of the spectrum and it has theadvantages of short wavelength and high output power. As a result, the excimer laserpromises a whole range of technological applications in areas of the semiconductor andchemical industries.

Today, optoelectronics covers an extensive range of applications. Light-based technolo-gies are now applied not only to optical communications but also to industrial processing,medicine, measurement and data processing. At present, there is a growth in the num-ber of companies offering a range of devices and systems. They cover the manufactureof materials and components for lasers, optical fibres, optical ICs and the production ofoptical measuring instruments. Optical processing and communication devices and opti-cal disks are based largely on laser light technologies and they may also include relatedoptical technologies developed from laser techniques.

1.3 Optical Fibre Communication Systems

By 1980, advances in optical technologies had led to the development and worldwideinstallation of practical and economically feasible optical fibre communication systemsthat carry live telephone, cable TV and various types of telecommunications traffic. Theseinstallations all operate as baseband systems in which the data is sent by simply turningthe transmitter on and off. Despite their apparent simplicity such systems have alreadyoffered very good solutions to some vexing problems in conventional applications.


Data communication capacity and the distances over which the data can be transmittedhave continued to increase. Commercial long-distance fibre communication systems beinginstalled in the mid and late 80s use the ‘second-generation’ 1300 nm technology andtransmit data at 256 Mbps on each of several fibres over distances of 50 km betweenrepeaters. The first commercial long-distance optical fibre telephone system was put intoservice by the American Telephone and Telegraph Company (AT&T) in 1985 [18]. Today,most major cities in the USA, Europe, Japan, South Korea and some other countries arelinked by optical fibre systems. For example, in Japan, Nippon Telegraph and Telephonepublic corporation (NTT) has played an important role in the research and development ofoptical communications. NTT has completed development of small and medium-capacity(100 Mbps to 1 Gbps) as well as large-capacity (10, 40 and 100 Gbps) inland trunk andsubmarine non-repeater systems. In the small- and medium-capacity systems graded indexmultimode fibres – and in the large-capacity systems single-mode fibres – have been used.In other countries, optical communication systems have been introduced in on-land trunknetwork systems. In USA, between 1981 and 1985, 6, 45, 90 and 430 Mbps systems, and inthe UK, 8, 34 and 140 Mbps, and in France 34 and 140 Mbps have been used successfully.

The capacity of a communication system is often measured through the bit rate-distanceproduct B × L, where B is the bit rate and L is the repeater spacing. This producthas increased through technological advances since 1850 from 10 bps.km to beyond1 Tbs.km [19]. Also, the progress in the performance of lightwave systems has indeedbeen rapid, as evidenced by the several-orders-of-magnitude improvement in the bit rate-distance product over the recent period [19].

Optical transmission can be classified into short-distance and long-distance categoriesdepending on whether the optical signal is transmitted over relatively short or long dis-tances compared with typical intercity distances of 50–100 km. Short-haul optical systemscover intercity and local loop traffic, and typically operate at low bit rates over distancesof less than 20 km. While, long-haul fibre communication systems operate at high bitrates over long distances. A typical optical fibre system link is shown in Figure 1.2.It comprises three main parts:

• An optical transmitter consisting of an optical light source and its associated drivecircuitry;

• A transmission medium which is an optical fibre cable; and• An optical receiver consisting of a photodetector which is followed by an amplifier and

a signal demodulation circuit.

DataSignal

OpticalSource

Photo-detector

Optical Transmitter

Fibre Optic asCommunications Channel

Optical Receiver

DataDriver Circuitand Modulator

Amplifier and SignalDemodulation Circuit

Figure 1.2 Optical fibre communications system


The function of an optical transmitter is to convert the electrical signal into optical formand launch the resulting optical signal into the optical fibre. Both semiconductor laserdiodes (LD) and light emitting diodes (LED) are used as optical sources because of theircompatibility with optical fibres. The optical signal is produced by modulating the LD orLED optical carrier wave. Modulation can be performed either by an external modulatoror by direct modulation. In the latter method the input signal is applied to the driver circuitof the optical source like the one in Figure 1.2 and the signal is modulated by varying theinjection current of the semiconductor optical source. In the 0.7 μm to 0.9 μm wavelengthbands the light source materials are generally alloys of GaAlAs. At longer wavelengthsranging from 1.1 μm to 1.6 μm an InGaAsP alloy is the principal optical source material.

The optical signal is usually launched into the optical fibre via a micro-lens so as tomaximize the coupling efficiency between the optical source and the optical fibre [20].Often the launched power P0, is expressed in dBm units (i.e. 1 mW is 0 dBm) and is animportant design parameter as it determines how much fibre loss can be tolerated. Typicalvalues of P0 for LEDs are less than −10 dBm and for LDs are in the 0 to 20 dBm range.The LD or LED beam (i.e. the optical carrier) may be modulated using either an analogueor a digital information signal. Analogue modulation involves the variation of the lightemitted from the optical source in a continuous manner. With digital modulation, however,discrete changes in the light intensity are obtained similar to amplitude-shift keying (ASK)or on-off modulation (OOK) pulses. Although it is often easy to implement analoguemodulation with an optical fibre communication system, it is less efficient and requires afar higher signal-to-noise ratio at the receiver than digital modulation. Also the linearityneeded for analogue modulation is not always provided by semiconductor optical sources,especially at high modulation frequencies. So analogue optical fibre communication linksare generally limited to shorter distances and lower bandwidths than digital links.

The receiver in an optical fibre communication system essentially consists of the pho-todetector followed by an amplifier with possibly additional signal processing circuits.Therefore, the receiver initially converts the optical signal, incident on the detector intoan electrical signal, which is then amplified before further processing to extract the infor-mation originally carried by the optical signal, as shown in Figure 1.2. When an opticalsignal is launched into the optical fibre it becomes progressively attenuated and distortedwith increasing distance because of scattering, absorption and dispersion mechanisms inthe fibre waveguide. At the receiver the attenuated and distorted modulated optical poweremerging from the fibre will be detected by a photodetector (PD) and converted into anelectrical current output (i.e. photocurrent) and processed. There are two principal pho-todetectors used in a fibre-optic link, PIN and avalanche photodiodes (APD). Both devicesexhibit high efficiency and response speed. For applications in which a low-power opti-cal signal is received, an APD is normally used since it has a greater sensitivity owingto an inherent internal gain mechanism. Silicon PDs are used in the 0.7 μm to 0.9 μmregion whereas Germanium or InGaAs are the prime material candidates in the 1.1 μm to1.6 μm region.

The design of the optical receiver is inherently more complex than that of the transmittersince it has to both amplify and reshape the degraded signal received by the PD. Theprincipal figure of merit for a receiver’s sensitivity is the minimum optical power necessaryat the desired datarate to attain either a given error probability for digital systems or aspecified signal-to-noise ratio for analogue systems. The ability of a receiver to achieve


a certain performance level depends on the PD type, the noise performance in the systemand the characteristics of the successive amplification stages at the receiver.

In addition, there has been an ongoing research in soliton transmission [21]. A solitonis a non-dispersive pulse that makes use of nonlinearity in a fibre to cancel out chromaticdispersion effects. Researchers at NTT have transmitted solitons at 20 Gbps over 70 kmof optical fibre having on the average 3.6 ps/km/nm of dispersion [22]. In 1992, AT&Treported 32 Gbps optical soliton data transmission over 90 km [23], and also KDD reportedtransmission of 5 Gbps modulated optical solitons over 3000 km of optical fibre usingerbium-doped fibre amplifiers [24]. Much longer transmission distances have also beenreported using dispersion-shifted fibre [25].

1.4 Lightwave Technology in Future

Beyond the technologies on the immediate horizon and the systems proposed or envis-aged, several other opportunities can be dimly seen and still more are expected. In otherwords, despite this rapid growth and the many successful applications, lightwave tech-nology is not even close to being mature. New fibre materials and structures can greatlyincrease the available bandwidth and decrease the attenuation. Coherent communicationtechniques can improve receiver sensitivity and make more sophisticated and powerfulmodulation methods available to increase the system throughput such as multidimensionaland multilevel modulation techniques.

An important objective is the perfection of all-optical networks, which include all-optical signal processing, switches, repeaters and network units. Applications includelocal area networks (LANs), subscriber loops and TV distribution [26].

1.5 Optical Lightwave Spectrum

Light is electromagnetic energy which travels in a wavefront. Thus, it can be identifiedwithin the general electromagnetic spectrum of alternating waves. Light as an electromag-netic wave is characterized by a combination of time varying electric E and magnetic Hfields propagating through space as seen in Figure 1.3. The frequency of oscillation ofthe fields, f and their wavelength, λ are related:

f = c

nλ(1.1)

where c = 3 × 108 m/s is the light velocity in vacuum and n is the refractive index of themedium given by:

n = √εr · μr (1.2)

where εr , μr are the relative permeability and relative permittivity of the medium, respec-tively. The electric and magnetic fields oscillate perpendicularly to one another and also tothe direction of propagation as illustrated in Figure 1.3. The simplest waves are sinusoidalwaves, which can be expressed as:

E (z , t) = Eo cos(ωt − kz + �) (1.3)


Electric Field(E)

zλ

Magnetic Field(H)

Figure 1.3 E and H propagating in orthogonal planes, perpendicular to the direction ofpropagation

where E (z , t) is the value of the electric field at the point z and at time t , Eo is the ampli-tude of the wave, ω = 2π f is the angular frequency, k = 2π/λ is the wave number and� is the phase constant. The term (ωt − kz + �) is the phase of the wave. Equation (1.3)describes a perfectly monochromatic plane wave of infinite extent propagating in thepositive z direction.

To obtain a better understanding of the wave characteristics, Equation (1.3) can berepresented diagrammatically at one instance in time and for one point in space as shownin Figure 1.4(a). It is shown as the variation of the electric field, E with distance at t = 0,assuming � = 0. This spatial variation of the electric field is given by:

E = Eo cos (kz ) (1.4)

Similarly Figure 1.4(b) depicts the variation of the electric field as a function of time atsome specific location in space (e.g. z = 0) which is:

E = Eo cos (ωt) = Eo cos (2π ft) = Eo cos

(2π

Tt

)(1.5)

where T = 1/f is the period. The frequencies and the wavelengths of the electromagneticwaves as seen in Equation (1.3) are related by Equation (1.1).

The term ‘optical’ usually refers to frequencies in the infrared, visible and ultravioletportions of the electromagnetic spectrum. Figure 1.5 shows the electromagnetic spectrumrange of optical frequencies that primarily interest us. In this region it is customaryto specify the band of interest in terms of wavelength instead of frequency as in theradio region. The optical spectrum ranges from about 0.2 μm (the far ultraviolet) to about100 μm (the far infrared). Visible wavelengths extend from 0.4 μm (colour blue) to 0.7 μm(colour red) as shown in Figure 1.5. Optical fibres are not very good transmitters of light inthe visible region. They attenuate the waves to such an extent that only short transmissionlinks are practical. Losses in the ultraviolet are even greater. In the infrared, however, there


0 1 2 3 4

(a) (b)

5 6 7 8 z

E

t = 0 λE0

−E0

0 1 2 3 4 5 6 7 8 t

z

z = 0 TE0

−E0

Figure 1.4 Electric field E plotted as a function of (a) spatial coordinate z (b) time

Wavelength

Frequency

Radio waves Radar Infrared Ultraviolet X-radiation

RED

700

nm 600

nm 500

nm 400

nm

Crim

son

Yel

low

Cya

n

Vio

let

GREEN BLUE

Visible spectrum

FM radioand TV

Microwaves OpticalIR

OpticalUV

Gammaradiation

100

kHz

1 M

Hz

10 M

Hz

100

MH

z

1 G

Hz

10 G

Hz

100

GH

z

1 T

Hz

10 T

Hz

100

TH

z

1 P

Hz

10 P

Hz

100

PH

z

1 E

Hz

10 E

Hz

100

EH

z

1 km

10 k

m

100

m

10 m

1 m

100

mm

10 m

m

1 m

m

100

μm

10 μ

m

1 μm

100

nm

10 n

m

1 nm

100

pm

10 p

m

1 pm

Figure 1.5 Electromagnetic spectrum

are two regions in which glass is a very efficient transmitter. This occurs at wavelengthsclose to 0.85 μm and in the region from 1.1 to 1.6 μm as shown from Figure 1.1.

1.6 Optical Fibre Transmission

As the transmission medium in a communication system, optical fibres have the potentialfor being used wherever twisted wire pairs or coaxial cables are employed. This is dueto their lowloss and widebandwidth as apparent from Figure 1.1. Optical fibres have a


flat transfer function far beyond 100 MHz and have very low loss compared with wirepairs or coaxial cables. In short, communication using an optical carrier waveguide alonga glass fibre has the following extremely attractive features [11, 28–31]:

• Enormous potential bandwidth• Small size and weight• Electrical isolation• Immunity to interference and crosstalk• Signal security• Low transmission loss• Ruggedness and flexibility• System reliability and ease of maintenance• Potential low cost

The fundamental principles underlying these enhanced performance characteristics, andtheir practical realizations are considered as a whole medium in the systematic and net-working techniques in the following chapters. It is assumed that the readers of this bookhave general understanding of the basic principles and properties of light as well asfibre-optics.

The transport network, also known as the first-mile network, connects the serviceprovider central offices to businesses and residential subscribers. This network is alsoreferred to in the literatures as the ‘subscriber access network’, the ‘local loop’ or evensometimes the ‘last-mile’. Residential subscribers demand first-mile access solutions thathave high bandwidth, offer media-rich Internet services, and are comparable in price withexisting networks. Similarly, corporate users demand broadband infrastructure throughwhich they can connect their local area networks (LAN) to the Internet backbone.

1.7 Multiple Access Techniques

In order to make full use of the available bandwidth in optical fibres and to satisfythe bandwidth demand in future information networks, it is necessary to multiplex low-rate data streams onto optical fibre to increase the total throughput. There is a need fortechnologies that allow multiple users to share the same frequency, especially as wirelesstelecommunications continues to increase in popularity. Currently, there are three commontypes of multiple access systems:

• Wavelength division multiple access (WDMA)• Time division multiple access (TDMA)• Code division multiple access (CDMA)

This section reviews the basic multiple access techniques in the optical domain andintroduces the current state of optical access techniques.

1.7.1 Wavelength Division Multiple Access (WDMA)

In WDMA systems, each channel occupies a narrow optical bandwidth (≥100 GHz)around a centre wavelength or frequency [32].


Time, t

User 1

User 2

User j

λ1

λ2

λj

Wav

elen

gth,

λ

Figure 1.6 Resource sharing based on the WDMA technique

The modulation format and speed at each wavelength can be independent of those ofother channels as shown in Figure 1.6.

Arrayed or tuneable lasers will be needed for WDMA applications [33]. Because eachchannel is transmitted at a different wavelength, they can be selected using an opticalfilter [34]. Tuneable filters can be realized using acousto-optics [35], liquid crystal [36]or fibre Bragg grating [37]. To increase the capacity of the fibre link using WDMA weneed to use more carriers or wavelengths, and this requires optical amplifiers [38] andfilters to operate over extended wavelength ranges. Due to the greater number of channelsand larger optical power the increased nonlinear effects in fibres causes optical crosstalksuch as four-wave mixing [39] over wide spectral ranges. Another approach to increasethe capacity of WDMA links is to use dense WDM (DWDM) which will have to operatewith reduced channel spacing, and this is based on the ITU-T recommendation G.692 [40]defining 43 wavelength channels from 1530 to 1565 nm, with a spacing of 100 GHz asshown in Figure 1.7.

This requires a sharp optical filter with linear phase response, wavelength stable com-ponents and optical amplifiers with flat gain over wide bandwidths. Also, optical fibresmust support hundreds of channels without distortion or crosstalk. With respect to channel

1530 1535 1540 1545 1550 1555 1560 1565

Wavelength (nm)

100 GHz

Figure 1.7 The ITU-T DWDM spacing


switching, wavelength routing is the next switching dimension for DWDM, with interfer-ometric crosstalk being an essential issue in the implementation of cross-connects basedon space and wavelength [41]. Hence, the extent of wavelength routing that is realizableplaces fundamental limits on network flexibility, which in turn determines switch size,implementation complexity and costs.

1.7.2 Time Division Multiple Access (TDMA)

In TDMA system, each channel occupies a pre-assigned time-slot, which interleaves withthe time-slots of other channels as shown in Figure 1.8.

Synchronous digital hierarchy/synchronous optical network (SDH/SONET) is the cur-rent transmission and multiplexing standard for high-speed signals, which is based ontime division multiplexing [42]. Optical TDMA (OTDMA) networks can be based on abroadcast topology or incorporate optical switching [43]. In broadcast networks, there isno routing or switching within the network. Switching occurs only at the periphery ofthe network by means of tuneable transmitters and receivers. The switch-based networksperform switching functions optically within the network in order to provide packet-switched services at very high bit rates [44]. In an electrically time-multiplexed system,multiplexing is carried out in the electrical domain, before the electrical-to-optical (E/O)conversion and demultiplexing is performed after optical-to-electrical (O/E) conversion.Major electronic bottlenecks occur in the multiplexer E/O, and the demultiplexer O/E,where electronics must operate at the full multiplexed bit rate. As an alternate solution,the bottleneck of O/E/O conversion moved further in processing rather than right after themux/demux via optically time division multiplexed technique where the optical signalsremain in optical domain and push the E/O and O/E to the transmitters and receiversrespectively instead of at the mux/demux stage. OTDMA systems offer a large number ofnode addresses; however, the performance of OTDMA systems is ultimately limited by

Time, t

Use

r 2

tjt2t1

Wav

elen

gth,

λ

Use

r 1

Use

r j

Figure 1.8 Resource sharing based on the TDMA technique


the time-serial nature of the technology. OTDMA systems also require strong centralizedcontrol to allocate time-slots and to manage the network operation.

1.7.3 Code Division Multiple Access (CDMA)

CDMA is one of a family of transmission techniques generically called spread spectrum,explained in the following section. In this technique, the network resources are sharedamong users that are assigned a code instead of a time-slot as in TDMA or a wavelengthas in WDMA. Then, different users are capable of accessing the resources using the samechannel at the same time, as shown in the Figure 1.9.

The concepts of spread spectrum, i.e. CDMA, seem to contradict normal intuition,since in most communications systems we try to maximize the amount of useful signalwe can fit into a minimal bandwidth. In CDMA we transmit multiple signals over the samefrequency band, using the same modulation techniques at the same time [45]. Traditionalthinking would suggest that communication would not be possible in this environment.The following effects of spreading are worth mentioning.

1.7.3.1 Capacity gain

Using the Shannon–Hartly law for the capacity of a band-limited channel, it is easy tosee that, for a given signal power, the wider the bandwidth used, the greater the channelcapacity. So if we broaden the spectrum of a given signal we get an increase in channelcapacity and/or an improvement in the signal-to-noise ratio (SNR). The Shannon–Hartlylaw gives the capacity of a band-limited communications channel in the presence ofGaussian noise (most communications channel has Gaussian noise) [46].

Capacity = B · log

(1 + Ps

2BNo

)(1.6)

where Ps represents signal power, No noise power and B available bandwidth.

Time, t

User 2, Code 2

User 1, Code 1

User jCode j

Code, C

Wavelength, λ

Figure 1.9 Resource sharing based on the CDMA technique


It is easy to see that with Ps and No held constant, capacity increases as bandwidthincreases (though not quite as fast). Thus, for a given channel capacity, the required powerdecreases as utilized bandwidth increases. The wider the bandwidth the lower the powerthat we need to use for a given capacity.

1.7.3.2 Security

Spread spectrum was invented by military communications people for the purpose ofbattlefield communications. Spread spectrum signals have an excellent rejection ofintentional jamming (jammer power must be very great to be successful). In addition,the direct sequence (DS) technique results in a signal which is very hard to distinguishfrom background noise unless you know the particular random code sequence used togenerate the signal. Thus, not only are DS signals hard to jam, but they are extremelydifficult to decode (unless you have the key) and quite hard to intercept even if all youneed to know is when something is being transmitted.

1.8 Spread Spectrum Communications Techniques

Spread spectrum communication (SSC) involves spreading the desired signal over a band-width much larger than the minimum bandwidth necessary to send the signal. It hasbecome very popular in the realm of personal communications [47]. Spread spectrummethods can be combined with CDMA methods to create multiuser communicationssystems with very good interference performance.

This section covers the details behind the method of SSC, as well as analysing two maintypes of SS systems, namely direct-sequence spread spectrum (DS-SS) and frequency-hopping spread spectrum (FH-SS).

As stated above, spread spectrum systems afford protection against jamming and inter-ference from other users in the same band, as well as noise, by spreading the signalto be transmitted and performing the reverse de-spread operation on the received signalat the receiver. This de-spreading operation in turn spreads those signals which are notproperly spread when transmitted, decreasing the effect that spurious signals will haveon the desired signal. Spread spectrum systems can be thought of having two generalproperties: first, they spread the desired signal over a bandwidth much larger than theminimum bandwidth needed to send the signal, and second, this spreading is carried outusing a pseudo-random noise (PN) sequence. In general, we will see that the increase inbandwidth above the minimum bandwidth in a spread spectrum system can be thoughtof as applying gain to the desired signal with respect to the undesirable signals. We cannow define the processing gain GP as [48]:

GP = BWcar .

BWdata(1.7)

where BWcar . is the bandwidth that the signal has been increased to (i.e. carrier bandwidth),and BWdata is the minimum bandwidth necessary to transmit the data signal. Processinggain can be thought of as the improvement over conventional communication schemesdue to the spreading of the signal. Often, a better measure of this gain is given by thejamming margin, Mj :

Mj (dB) = GP (dB) − SNRmin (1.8)


This indicates the amount of interference protection offered before the signal is corrupted.The spreading function is achieved through the use of a PN sequence. The data signalis combined with the PN sequence such that each data bit is encoded with several if notall the bits in the PN sequence. In order to achieve the same data rate as was desiredbefore spreading, the new data must be sent at a rate equal to the original rate multipliedby the number of PN sequence bits (i.e. code-length) used to encode each bit of data.This increase in bandwidth is the ‘processing gain’, which is a measure of the noise andinterference immunity of this method of transmission.

To see how the spreading process helps protect the signal from outside interference, thetypes of possible interference are introduced: (i) noise, and (ii) interference from otherusers of the same frequency band. Noise can be considered as background additive whiteGaussian noise (AWGN) having power spectral density of N0. Since the noise is white (i.e.includes all frequencies) the spreading of the bandwidth does not have much of an effecthere. The noise power is constant over the entire bandwidth, thus increasing the bandwidthin fact lets more noise into the system, which might be seen as unfavourable. However,we will see that this is not really a problem. The major source of signal corruption comesfrom multiple user or multiple access interference (MAI). The technique of CDMA is tocombat this type of interference.

In a wireless communications network, all the signals propagate through the air byelectromagnetic waves, thus there is no way to ensure that a given user can receive theirdesired signal. In fact, the user receives all the signals that are sent in that band.

1.8.1 Direct-Sequence Spread Spectrum (DS-SS)

DS-SS is the most common version of spread spectrum in use today due to its simplicityand ease of implementation. In DS-SS, the carrier (data signal) is modulated by the PNsequence, which is of a much higher frequency than the desired data rate.

Let f be the frequency of the data signal, with an appropriate pulse period T = 1/f .Let the PN sequence be transmitted at a rate fc so that the increase in the data rate is fc/f .The frequency fc is known as the chip-rate, with each individual bit in the modulatingsequence known as a ‘chip’.

Thus the width of each pulse in the modulating sequence is Tc , the chip duration.Figure 1.10 illustrates the encoded signal, the data signal for one pulse width, and the PNsequence over the same time [49].

As a result, the frequency domain will look like the signal in Figure 1.11. Assumingthat the data signal is D(t), transmitted at frequency f , and the PN sequence is PN(t) atfrequency fc , then the transmitted signal is:

S (t) = D(t) · PN(t) (1.9)

The PN sequence is designed such that it has very good correlation properties, forexample:

RPN(τ ) ={

1 τ = 0, N , 2N

1/N otherwise(1.10)

where N is the length of the PN sequence (code-length).


Time, t

Data × PN

PN

Data 1

T

Tc

10 01

−1

1

0

1

0

Figure 1.10 DS-SS signalling format

Data Signal

Data SignalModulated with PN

fc−fc−1/T 1/T

Figure 1.11 Data signal and DS-SS modulated data signal in the frequency domain

Therefore, when the signal is correlated with the PN sequence at the receiver, thereceived signal will be recovered exactly (assuming that there is synchronization betweenthe sent and received PN sequences) as:

S (t) · PN(t) = D(t) · PN(t) · PN(t) = D(t) (1.11)

Here we recall the correlation functions. Two signals are correlated by multiplying eachother bit by bit in discrete format. When a signal is correlated by its own shifted version,for example, let xi be the sequence code, xi+τ be the signal shifted by time τ , and n bethe PN code-length, then the auto-correlation function is:

RXX (τ ) =n−1∑i=0

xi xi+τ (1.12)


When two signals xi and yi are correlated the cross-correlation function is:

RXY (τ ) =n−1∑i=0

xi yi+τ (1.13)

where yi+τ is the signal shifted by time τ . It should be noted that two signals are ‘orthog-onal’ when the cross-correlation value between them is zero. If we allow both noise anda jamming signal J (t) with finite power distributed evenly across the frequency band, thereceived signal at the input to the receiver, Y (t) is:

Y (t) = D(t) · PN (t) + J (t) + N (t) (1.14)

Now, this received signal is correlated with the associated PN sequence at the receiver.The result of correlating Y (t) and the PN sequence is the product of PN (i.e. PN (t))with D(t) · PN (t), J (t) and N (t). Thus, the product of D(t) · PN (t) · PN (t) equals D(t)which is the data signal, since the PN (t) · PN (t) represents the auto-correlation andde-spreads the data signal back to the original frequency of f = 1/T . While, the productof J (t) · PN (t) and N (t) · PN (t) denotes the further spreading the jamming and noisesignals respectively. Therefore, the jamming and noise signals power reduces within thecarrier frequency of fc . Using a matched filter at the receiver to pass the data signal resultsin a decrease in the jamming and noise power by a factor of the processing gain GP i.e.GP = BWCar .

BWdata≈ fc/f .

So we see that the data signal has been made immune to the effect of a maliciousthird-party jammer as well. As stated earlier, even though a factor of fc/f more noise waslet into the system by the increased bandwidth, the effect of that noise was also reducedby GP due to the processing gain of the system, and thus the effect of AWGN has notbeen increased by this DS-SS system [48].

1.8.2 CDMA and DS-SS

In a CDMA system, each user is identified by its own spreading code, and in order toprevent users from interfering with each other, these codes are designed to be orthogonalto each other (ideally, the cross-correlation function between any two of these codes iszero). In practice, perfect orthogonality is difficult to achieve, but for now we assumeperfect orthogonality in order to easily understand the concept of CDMA theory. Eachuser’s signal is being encoded with not only a PN sequence, but also its own orthogonalcode. Therefore, the transmitted signal S (t) is:

Si (t) = Di (t) · PN(t) · Ci (t) = Di (t) · Pi (t) (1.15)

where Ci (t) denotes the CDMA code of the i th user whose data signal is Di (t), andPi (t) denotes the combination of the PN sequence and the orthogonal code for the i th

user. Ideally, this allows a large number of users to use the same bandwidth; that is, nownot only we do have the intentional interference rejection properties but also we have a


Data Signal

Demodulator(to Baseband Freq.)

X X

(a)

(b)

Modulator(to Carrier Freq.)

DecisionDevice

Ci(t)

Ci(t)

PN(t)

PN(t)

X X Integrator

Figure 1.12 DS-SS basic transceiver (a) transmitter (b) receiver

multiuser interference rejection. Assume that there are N users with N orthogonal codesin this system, all using the same frequency band. Thus the i th receiver’s signal is:

Yi (t) = Di (t) · P(t) +N∑

k=1,k �=i

Dk (t + θ) · Pk (t + θ) (1.16)

where θ is a random delay. When this is correlated with the PN sequence and the i th

orthogonal code, the result will become zero (i.e. orthogonality), and only the signal dueto the desired transmission will remain. The basic transmitter and receiver structures forDS-SS are shown in Figure 1.12. The transmitter only multiplies the data signal withthe PN sequence and the CDMA code, and then modulates the resulting signal to thecarrier frequency, and the receiver just performs the reverse operation and integrates thereceived signal. However, all this assumes perfect synchronization between transmitterand receiver.

1.8.3 Frequency-Hopping Spread Spectrum (FH-SS)

In FH-SS, the signal itself is not spread across the entire large bandwidth, rather the widebandwidth is divided into N sub-bands, and the signal hops from one band to the otherin a pseudo-random manner instead. The centre frequency of the signal changes fromone sub-band to another, as shown in Figure 1.13. As one can see, the large frequencybandwidth N · fb at fc is divided into N sub-bands of width fb . The bandwidth fb must beenough to transmit the data signal D(t), and, at a predetermined time interval, the centrefrequency of the data signal changes from one sub-band to another pseudo-randomly [47].

As Figure 1.13 shows, the data signal hops from band N at fc + (N /2)fb to band 2 atfc − ((N /2) − 1)fb and to band N − 2, and so on. Usually, the width of each sub-bandis set so that the amount of signal that overlaps with adjacent sub-bands is minimal, andis thus approximately the bandwidth of the original data signal.


Band 1

fc − (N/2)fb fc+ (N/2)fb

f

fc − (N/2−1)fb fc+ (N/2 −1)fb

Band 2 Band 3BandN− 2

BandN−1 Band N

Hop #H+1

Hop #H−1Hop #H

Figure 1.13 FH-SS signalling format

Two different kinds of FH-SS are commonly used: slow FH and fast FH. In slowFH-SS, several bits are sent for each hop, so the signal stays in a particular sub-band fora long time relative to the data-rate. In fast FH-SS, the signal switches sub-bands severaltimes for each bit transmitted, so the signal stays in a sub-band for a very short timerelative to the data-rate. It should be noted that slow FH is not really a spread spectrumtechnique, since this does not really spread the system and also, because the time spentin one sub-band is very large, the corresponding width of the band can be small, thuspossibly violating the first principle of a spread spectrum system, namely that the spreadbandwidth must be much greater than the non-spread bandwidth.

In fast FH, the performance of the system with respect to AWGN is not affected asin the direct sequence (DS). The noise power seen at the receiver is approximately thesame as that in the un-hopped case, since each sub-band is approximately the same sizeas the original data signal’s bandwidth. Here, if we again assume that the jamming signalJ (t) is distributed uniformly over the entire band, it is clear that the only portion of thejamming signal that affects the data is the part within the bandwidth of fb , and thus thejamming signal is reduced by the factor of the processing gain GP which is defined as:

GP = BWCar .

BWdata= N · fb

fb= N (1.17)

Thus in FH, the protection afforded is equal to the number of frequency bands. In caseof interference in certain frequency bands, the probability of a bit being in error is thengiven by PBE = J /N , where J is the number of channels interfered, and N is the totalnumber of frequencies available to the hopping.

However, FH allows us to very simply decrease the bit-error rate (BER). If we chooseto have a large number of chips per bit (i.e. a chip represents a hop), then we can use asimple majority function to determine what the transmitted bit was. We assume that thenumber of available hop channels is larger than the number of channels being interfered.If the simple majority function is being used, then the formula for the error rate becomes:

PE =c∑

x=r

(cx

)px · qc−x (1.18)

where

(cr

)= c!/[(c − r)! · r!] and it reads the combination of r out of c which is

the number of chips per bit (hops per bit), r is the number of chip errors necessaryto cause a bit error, p is the probability of one bit error (i.e. PBE = J /N ), and q is the


DataSignal

DecisionDevice

BPFX

Channel

X

PN SequenceGenerator

PN SequenceGenerator

FrequencySynthesizer

(a) (b)

FrequencySynthesizer

Figure 1.14 FH-SS basic transceiver (a) transmitter (b) receiver

probability of no error for a chip, or 1 − p. By increasing the number of chips per bit fromone to three and assuming r = 2 and p = 0.01, for example, we find that the error ratewill be:

PE =c∑

x=r

(cx

)px · qc−x =

(32

)(p2 − p3) +

(33

)p3

= 3p2 − 2p3 = 3 × 10−4 − 2 × 10−6 ≈ 3 × 10−4 (1.19)

Thus by just increasing the hopping rate from once per bit to three times per bit, thebit-error rate can be decreased dramatically.

The PN sequence is used here to determine the hopping sequence. So in order totransmit the signal, the data signal must be modulated up to the centre frequency of theband determined by the PN sequence. The structure of the transmitter is as shown inFigure 1.14. The data signal is modulated by the PN sequence-generator frequency by thefrequency synthesizer and transmitted. The frequency synthesizer demodulates the signaldown to a baseband frequency, and then the signal is filtered – thus only the desired datasignal is passed through – and finally the signal is decoded. Again, to get multiple usersusing the same wide frequency band, CDMA techniques must be used.

1.8.4 CDMA and FH-SS

The method of CDMA used in this case is to provide each user with an orthogonal hopsequence, i.e. no two users occupy the same sub-band at the same time. In this way,multiple users can be accommodated without any chance of interfering with one another,since ideally only one user will be in a frequency sub-band during a given hop, andthus the receiver, due to its band-pass filter, will be able to detect the intended signal.Thus the transceiver given in Figure 1.14 needs only to be modified to incorporate anorthogonal code sequence in determining the centre frequency of the current hop (i.e.use the orthogonal code combined with the PN sequence as the input to the frequencysynthesizer), and the system can support multiple users.


1.9 Motivations for Optical CDMA Communications

Multiple access techniques are required to meet the demand for high-speed and large-capacity communications in optical networks, which allow multiple users to share thefibre bandwidth. Three major multiple access techniques have been looked at above: eachuser is allocated a specific time-slot in time-division multiple-access (TDMA), or a spe-cific frequency (wavelength) slot in wavelength division multiple-access (WDMA). Bothtechniques have been extensively explored and utilized in optical communication systems[32–34, 36, 40, 50–58]. The alternative, optical code-division multiple-access (OCDMA)[59–80], is receiving increasing attention due to its potential for enhanced informationsecurity, simplified and decentralized network control, improved spectral efficiency andincreased flexibility in the granularity of bandwidth that can be provided. In optical CDMA(OCDMA), different users whose signals may be overlapped both in time and frequencyshare a common communications medium; multiple-access is achieved by assigning unlikeminimally interfering code sequences to different transmitters, which must subsequentlybe detected in the presence of multiple access interference (MAI) from other users.

CDMA is derived from radio frequency (RF) spread spectrum communications, origi-nally developed for military applications due to an inherent low probability of interceptand to immunity to interference, and more recently for commercial cellular radio applica-tions such as third-generation (3G) and beyond [47–49, 81]. CDMA is now becoming thedominant multiple-access technique in RF wireless networks. In contrast, the need to per-form encoding and decoding for optical CDMA (OCDMA) poses one immediate challengebecause of both the optical carrier frequency and the much higher bit rate of multi-gigabit/sper user, which already approaches the limit of electronic processing. Furthermore, thechallenges for OCDMA come from critical requirements which are routinely needed inoptical communication systems. Therefore, innovative all-optical processing technologiesare also wanted. These requirements include: extreme high quality-of-service (QoS), i.e.BER at 10−9 or below; large capacity, i.e. hundreds of users; enhanced bandwidth of100+ Gbps; and higher scalability at long distances such as tens or hundreds of kilome-tres for LANs and metropolitan area networks (MAN). Significant progress of OCDMAresearch has been achieved worldwide in recent years, including several different OCDMAschemes that have been proposed based on different choices of sources [54, 66, 70, 82],coding schemes [37, 59, 67] and detections [76, 80, 83–87] which will be discussed indetail in this book through the following chapters.

OCDMA techniques may be classified according to the choice of coherent versus inco-herent processing; coherent versus incoherent broadband optical source; and encodingmethods, i.e. time versus frequency domain and/or amplitude versus phase spectra. Toincrease the available coding space, time-wavelength (two-dimensional) coding schemeshave been proposed [88–91] where each code chip corresponds to a specific time posi-tion and wavelength position within a bit, as determined by a code matrix. However, thisbrings considerable complexity to the system implementations. Various spreading codingtechniques will be discussed in detail in Chapter 2.

CDMA is well suited for bursty network environments, and the asynchronous natureof data transmission can simplify and decentralize network management and control.However, due to complex system requirements, full asynchronism is very difficult to


implement in practice while real-time simultaneous MAI suppression, due to imperfectspreading codes, is still a hot research topic and under investigation [70, 72, 75, 79, 86,92–96]. On the other hand, the synchronous scheme takes advantage of accommodatinghigher numbers of subscribers due to the time-shift property of the codes, and althoughthe synchronization is difficult some methods have been proposed and are in use [97–99].

Several challenging research topics are still missing for practical OCDMA realizationand development. These include the high co-channel interference (i.e. MAI) naturallypresent in almost all forms of OCDMA; low network capacity in terms of number ofconcurrent users; and codes that can support various traffic demands in terms of bandwidthand BER performance.

The proposed improvement in channel utilization and relatively low technical complex-ity and ease of implementation will have a direct impact on the current state of OCDMAnetworking, as will be discussed in Chapter 9 concerning scalability and compatibilityof OCDMA with Internet protocol (IP) and passive optical networks (PON). A few sig-nificant CDMA properties of which simple, very high speed and cost-effective opticaltransport network can take advantage are as follows [71].

• Statistical allocation of network capacity : Any particular CDMA receiver experiencesother users’ signals as noise. This means that you can continue adding channels untilthe signal-to-noise ratio (SNR) becomes low enough that you start getting bit errors. Alink can be allocated as many active connections such that the total data traffic staysbelow the channel capacity. For example, if there are a few hundred voice channels overCDMA, and the average power is the channel limit, then many more voice connectionscan be managed than with TDMA or WDMA methods. This can also be applied to datatraffic on a LAN or access network where the traffic is inherently bursty in nature, aswith Internet protocol (IP).

• No guard time or guard bands: In a TDMA system when multiple users share thesame channel there must be a way to ensure that they do not transmit at the same timeand overlap each other’s signals. Since there is no really accurate clock recovery, alength of time must be allowed between the end of one user’s transmission and thebeginning of the next. As the speed gets higher, this guard time comes to dominate thesystem throughput. In a CDMA network, the stations simply transmit when they areready. Also, in a WDMA system, unused frequency space is allocated between bandsdue to frequency overlapping avoidance through filtering process. These guard bandsagain represent wasted bandwidth.

• Easier system management : The users must have frequencies and/or time-slots assignedto them through some central administration processes in WDMA and TDMA systems.All you need with CDMA is for communicating stations to have the assigned code.

1.10 Access Networks Challenges

High capacity backbone networks have been highlighted. The high capacity standardOC-192 backbone networks are currently provided by operators with 10 Gbps data rate.State-of-the-art access network technologies such as digital subscriber line (DSL) provide


up to a few Mbps, at best, of downstream bandwidth to customers, depending on variousflavours. Therefore, the access network in fact challenges the current technology ofnetworking on how to transfer a huge amount of traffic from backbone (i.e. core net-work) to access network (i.e. end-users) through wide bandwidth, high data rate services,such as video-on-demand (VoD) [100].

On the other hand, DSL technologies are highly distance-dependent in that the dis-tance of any DSL subscriber to a central office must be less than 6 km due to the signalattenuation and distortion. And, usually, service providers are reluctant to service todistances more than 4 km. However, different DSL flavours such as very high bit rateDSL (VDSL) can support up to 50 Mbps of downstream bandwidth which is emerging.It should be noted that when the data rate goes high on DSL technologies the dis-tances also shrink: for example, the maximum distance which VDSL is supported is only500 metres [101].

Alternatively, community access television (CATV) networks are commonly used forbroadband access [43]. By dedicating some radio frequency (RF) channels in coaxialcable for data, CATV networks provide Internet services as well as TV channels. Bearingin mind that CATV networks are mainly focused on broadcast services (like TV), theyhardly fit well in distributed access bandwidth. This could be the reason why CATVend-users become frustrated at higher loads. There is now a strong feeling that (i) faster(ii) more reliable and (iii) higher scalable access networks are needed for next-generationultra-high-speed crystal-clear broadband applications (e.g. high-definition IPTV).

The most promising proposed solution is to bring fibre closer to the home (i.e. end-users)in the next-generation access networks. The proposed models of FTTx [102, 103] includ-ing: fibre-to-the-home (FTTH), fibre-to-the-curb (FTTC), fibre-to-the-building (FTTB),etc. potentially offer extensive access bandwidth to their subscribers. FTTx aims to takeover technologies such as VDSL by providing fibre directly to the homes, or very closeby. FTTx platforms are mainly based on the passive optical network (PON) architectureswhich will be studied in detail in Chapter 9.

1.11 Summary

This chapter reviews the historical evolution of optical communications over fibre-optic inthe past 50 years since laser discovery. The need for high-speed high-bandwidth reliableand scalable communications networks has led the research communities and industryto develop transatlantic and transpacific fibre-optic communication links. The carry-onmotivations bring the optical fibre to homes and premises to serve the end-users withhigher quality services such as multimedia-on-demand and high-definition holographictelepresence.

To take advantages of extremely wide bandwidth offered by fibre-optics, severalmultiple access techniques are utilized to share the communication channel efficiently.Among them, the most popular ones, wavelength-division multiple access (WDMA),time-division multiple access (TDMA) and the more promising code-division multipleaccess (CDMA) techniques, have already been introduced. Spread-spectrum commu-nication methods including direct-sequence spread spectrum and frequency-hoppingspread spectrum have also been studied as a prerequisite to understanding CDMAprocedure. Chapter 2 is dedicated to introducing various types of spreading code designs


deployed in the optical domain for OCDMA, while later chapters are more dedicated tosystematic and networking aspects of OCDMA for various applications depending onthe specifications.

References1. Maiman, T.H. (1960) Stimulated optical radiation in ruby. Nature, 187, 493–494.2. Javan, A., Bennett Jr., W.R. and Herriot, D.R. (1961) Population inversion and continuous optical

MASER oscillation in a gas discharge containing a He-Ne mixture. Physical Review Letters , 6 (3),106–110.

3. Nathan, M.I. et al . (1962) Stimulated emission of radiation from GaAs p-n junction. Applied PhysicsLetters , 1 (3), 62–64.

4. Quist, T.M. et al . (1962) Semiconductor MASER of GaAs. Applied Physics Letters , 1 (4), 91–92.5. Holonyak Jr., N. and Bevacqua, S.F. (1962) Coherent (visible) light emission from GaAS1-xPx Junctions.

Applied Physics Letters , 1 (4), 82–83.6. Kompfner, R. (1965) Optical communications. Science, 150, 149–155.7. Kraemer, A.R. (1977) Free-space optical communications. Signal , 32, 26–32.8. Staff, L.F. (1980) Blue-green laser links to subs. Laser Focus , 16, 14–18.9. Kao, K.C. and Hockman, G.A. (1986) Dielectric fibre surface waveguides for optical frequencies. Proc.

IEE , 133 (3), 1151–1158.10. Kapron, F.P., Keck, D.B. and Maurer, R.D. (1970) Radiation losses in glass optical waveguides. Applied

Physics Letters , 17 (10), 423–425.11. Cherin, A.H. (1983) An introduction to optical fibres . McGraw-Hill, USA.12. Tran, D.C., Sigel Jr., G.H. and Bendow, B. (1984) Heavy metal fluoride glasses and fibres: A review.

J. Lightw. Technol., 2 (10), 566–586.13. Lucas, J. (1989) Review: fluoride glasses. J. Materials Science, 24 (1), 1–13.14. Folweiler, R.C. (1989) Fluoride glasses. GTE J. Science and Tech., 3 (1), 25–37.15. Ghafouri-Shiraz, H. and Shum, P. (1992) Simulation of the Erbium-doped fibre amplifier characteristics.

J. Microw. & Opt. Tech. Let., 5 (4), 191–194.16. Ghafouri-Shiraz, H. and Lo, B.S.K. (1996) Distributed feedback laser diodes: principles and physical

modelling . John Wiley and Sons, UK.17. Ghafouri-Shiraz, H. and Tsuji, S. (1994) Strain effects on refractive index and confinement factor of

InxGa(1-x)As laser diodes. J. Microw. & Opt. Tech. Let. (Special Issue), 7 (3), 113–119.18. O’Neill, E.F. (1985) A history of science and engineering in Bell systems . AT&T Bell Lab.19. Agrawal, G.P. (1992) Fibre-optic communication systems . John Wiley and Sons, USA.20. Kotsas, A., Ghafouri-Shiraz, H. and Maclean, T.S.M. (1991) Microlens fabrication on single-mode fibres

for efficient coupling from laser diodes. J. Optical and Quantum Electronics , 23 (3), 367–378.21. Hasegawa, A. (1989) Optical soliton in fibres . Springer-Verlag, Germany.22. Watsuki, K.T. et al . (1990) 20 Gb/s optical soliton data transmission over 70 km using distributed fibre

amplifiers. IEEE Photonics Tech. Letters , 2 (12), 905–907.23. Andrekson, P.A. et al . (1992) 32 Gb/s optical data transmission over 90 km. IEEE Photonics Tech. Letters ,

4 (1), 76–79.24. Taga, H. et al . (1992) 5 Gbits/s optical soliton transmission experiment over 3000 km employing 91

cascaded Er-doped fibre amplifier repeaters. Electronics Letters , 28 (24), 2247–2248.25. Nakazawa, M. et al . (1991) 10 Gbit/s soliton data transmission over one million kilometers. Electronics

Letters , 27 (14), 1270–1272.26. Horimatsu, T. and Sasaki, M. (1989) OEIC technology and its application to subscriber loops. J. Lightw.

Technol., 7 (11), 1612–1622.27. Electromagnetic Spectrum. Available from: http://www.lbl.gov/MicroWorlds/ALSTool/EMSpec/

EMSpec2.html (accessed 22 August 2011).28. Senior, J.M. (1992) Optical Fibre Communications Principles and Practice. Prentice-Hall (Second

Edition), Prentice-Hall Europe.29. Born, M., Wolf, E. and Bhatia, A.B. (1999) Principles of optics . 7th ed. Cambridge University Press,

New York, USA.


30. Zanger, H. and Zanger, C. (1992) Fiber optics communication and other applications . Macmillan, NewYork, USA.

31. Palais, J.C. (1988) Fiber optic communications . Prentice-Hall International Editions, New York.32. Borella, M.S. et al . (1997) Optical components for WDM lightwave networks. In: Proc. of IEEE , 85 (8).33. Lee, T.P. and Zah, C.E. (1989) Wavelength-tunable and single-frequency lasers for photonic communi-

cations networks. IEEE Comm. Mag., 27 (10), 42–52.34. Kobrinski, H. and Cheung, K.W. (1994) Wavelength-tunable optical filters: applications and technologies.

IEEE Comm. Mag., 32 (12), 50–54.35. Baron, J.E. et al . (1989) Multiple channel operation of an integrated acousto-optic tunable filter. Elec-

tronics Letters , 25 (6), 375–376.36. Sneh, A. and Johnson, K.M. (1994) High-speed tunable liquid crystal optical filter for WDM systems. In:

Proc. IEEE/LEOS Summer Topical Meetings on Optical Networks and Their Enabling Technologies .37. Ito, M. et al . (1995) Fabrication and application of fiber Bragg gratings review. J. Optoelectron. Devices

Technol., 10 (3), 119–130.38. Morkel, P.R. et al . (1991) Erbium-doped fiber amplifier with flattened gain spectrum. IEEE Photonics

Tech. Letters , 3 (2), 118–120.39. Chraplyvy, A.R. et al . (1994) Reduction of four-wave mixing crosstalk in WDM systems using unequally

spaced channels. IEEE Photonics Tech. Letters , 6 (6), 754–756.40. Brackett, C.A. (1990) Dense wavelength division multiplexing networks: principle and applications. IEEE

J. on Selected Areas in Comm., 8 (8), 948–964.41. Hinton, H.S. (1990) Photonic switching fabrics. IEEE Comm. Mag., 28 (4), 71–89.42. Perros, H.G. (2005) Connection-oriented networks: SONET/SDH, ATM, MPLS and optical networks . John

Wiley & Sons, Chichester, England.43. Ramaswami, R. and Sivarajan, K.N. (1998) Optical Networks: a practical perspective. Morgan Kaufmann.44. Mukherjee, B., Yao, S. and Dixit, S. (2000) Advances in photonic packet switching: an overview. IEEE

Comm. Mag., 38 (2), 84–94.45. Ilyas, M. and Moftah, H.T. (2003) Handbook of optical communication networks . CRC Press, Florida,

USA.46. Proakis, J.G. (1995) Digital communications . McGraw-Hill, New York, USA.47. Prasad, R. (1996) CDMA for wireless personal communications . Artech House publisher, Boston, USA.48. Meel, I.J. (1999) Spread spectrum – introduction and application . Siruis Communication, Malaysia.49. Viterbi, A.J. (1995) CDMA, principles of spreading spectrum communication. Addison Wesley, Boston,

USA.50. Eisenstein, G., Tucker, R.S. and Korotky, S.K. (1988) Optical time-division multiplexing for very high

bit rate transmission. J. Lightw. Technol., 6 (11), 1737–1749.51. Fujiwara, M. et al . (2002) Novel polarization scrambling technique for carrier-distributed WDM networks.

In: ECOC .52. Xu, R., Gong, Q. and Ya, P. (2001) A novel IP with MPLS over WDM-based broadband wavelength

switched IP network. J. Lightw. Technol., 19 (5), 596–602.53. Iwatsuki, K., Kani, J.I. and Suzuki, H. (2004) Access and metro networks based on WDM technologies.

J. Lightw. Technol., 22 (11), 2623–2630.54. Tsang, W.T. et al . (1993) Control of lasing wavelength in distributed feedback lasers by angling the active

stripe with respect to the grating. IEEE Photonics Tech. Letters , 5 (9), 978–980.55. Assi, C., Ye, Y. and Dixit, S. (2003) Dynamic bandwidth allocation for quality of service over Ethernet

PON. IEEE J. on Selected Areas in Comm., 21 (11), 1467–1477.56. Killat, U. (1996) Access to B-ISDN via PON – ATM communication in practice. Wiley Teubner

Communications, Chichester, England.57. Lam, C.F. (2007) Passive optical network: principles and practice. Academic Press, Elsevier, USA.58. Kramer, G. (2005) Ethernet passive optical network . McGraw-Hill, New York, USA.59. Azizoghlu, M., Salehi, J.A. and Li, Y. (1992) Optical CDMA via temporal codes. IEEE Trans on Comm.,

40 (8), 1162–1170.60. Heritage, J.P., Salehi, J.A. and Weiner, A.M. (1990) Coherent ultrashort light pulse code-division multiple

access communication systems. J. Lightw. Technol., 8 (3), 478–491.61. Salehi, J.A. (1989) Code division multiple-access techniques in optical fiber networks – part I: fundamen-

tal principles. IEEE Trans. on Comm., 37 (8), 824–833.


62. Salehi, J.A. and Brackett, C.A. (1989) Code division multiple-access technique in optical fiber networks –part II: system performance analysis. IEEE Trans. on Comm., 37 (8), 834–842.

63. Kwong, W.C., Perrier, P.A. and Prucnal, P.R. (1991) Performance comparison of asynchronous andsynchronous code-division multiple-access techniques for fiber-optic local area networks. IEEE Trans. onComm., 39 (11), 1625–1634.

64. Wei, Z. and Ghafouri-Shiraz, H. (2002) Proposal of a novel code for spectral amplitude coding opticalCDMA systems IEEE Photonics Tech. Letters , 14 (3), 414–416.

65. Smith, E.D.J., Blaikie, R.J. and Taylor, D.P. (1998) Performance enhancement of spectral-amplitude-coding optical CDMA using pulse position modulation. IEEE Trans. on Comm., 46 (9), 1176–1185.

66. Wei, Z., Ghafouri-Shiraz, H. and Shalaby, H.M.H. (2001) Performance analysis of optical spectral-amplitude-coding CDMA systems using super-fluorescent fiber source. IEEE Photonics Tech. Letters ,13 (8), 887–889.

67. Kavehrad, M. and Zaccarin, D. (1995) Optical code division-multiplexed systems based on spectral encod-ing of noncoherent sources. J. Lightw. Technol., 13 (3), 534–545.

68. Wei, Z. and Ghafouri-Shiraz, H. (2002) IP transmission over spectral-amplitude-coding CDMA links.J. Microw. & Opt. Tech. Let., 33 (2), 140–142.

69. Wei, Z. and Ghafouri-Shiraz, H. (2002) IP routing by an optical spectral-amplitude-coding CDMAnetwork. IEE Proc. Communications , 149 (5), 265–269.

70. Cooper, A.B. et al . (2007) High spectral efficiency phase diversity coherent optical CDMA with lowMAI. In: Lasers and Electro-Optics (CLEO).

71. Prucnal, P.R. (2005) Optical code division multiple access: fundamentals and Applications . CRC Taylor &Francis Group.

72. Shalaby, H.M.H. (1995) Synchronous fiber-optic CDMA systems with interference estimators. J. Lightw.Technol., 17 (11), 2268–2275.

73. Shalaby, H.M.H. (1995) Performance analysis of optical synchronous CDMA communication systemswith PPM signaling. IEEE Trans. on Comm., 43 (2/3/4), 624–634.

74. Shalaby, H.M.H. (1999) A performance analysis of optical overlapping PPM-CDMA communicationsystems. J. Lightw. Technol., 19 (2), 426–433.

75. Lee, T.S., Shalaby, H.M.H. and Ghafouri-Shiraz, H. (2001) Interference reduction in synchronous fiberoptical PPM-CDMA systems J. Microw. & Opt. Tech. Let., 30 (3), 202–205.

76. Shalaby, H.M.H. (1999) Direct-detection optical overlapping PPM-CDMA communication systems withdouble optical hard-limiters. J. Lightw. Technol., 17 (7), 1158–1165.

77. Hamarsheh, M.M.N., Shalaby, H.M.H. and Abdullah, M.K. (2005) Design and analysis of dynamiccode division multiple access communication system based on tunable optical filter. J. Lightw. Technol.,23 (12), 3959–3965.

78. Shalaby, H.M.H. (2002) Complexities, error probabilities and capacities of optical OOK-CDMA commu-nication systems. IEEE Trans on Comm., 50 (12), 2009–2017.

79. Shalaby, H.M.H. (1998) Cochannel interference reduction in optical PPM-CDMA systems. IEEE Trans.on Comm., 46 (6), 799–805.

80. Shalaby, H.M.H. (1998) Chip-level detection in optical code division multiple access. J. Lightw. Technol.,16 (6), 1077–1087.

81. Buehrer, R.M. (2006) Code division multiple access (CDMA). Morgan & Claypool Publishers, Colorado,USA.

82. Huang, W., Andonovic, I. and Tur, M. (1998) Decision-directed PLL for coherent optical pulse CDMAsystem in the presence of multiuser interference, laser phase noise and shot noise. J. Lightw. Technol.,16 (10), 1786–1794.

83. Liu, X. et al . (2004) Tolerance in-band coherent crosstalk of differetial phase-shift-keyed signal withbalanced detection and FEC. IEEE Photonics Tech. Letters , 16 (4), 1209–1211.

84. Betti, S., Marchis, G.D. and Iannone, E. (1992) Polarization modulated direct detection optical transmissionsystems. J. Lightw. Technol., 10 (12), 1985–1997.

85. Ohtsuki, T. (1999) Performance analysis of direct-detection optical CDMA systems with optical hard-limiter using equal-weight orthogonal signaling. IEICE Trans. on Comm., E82-B (3), 512–520.

86. Wang, X. et al . (2006) Demonstration of DPSK-OCDMA with balanced detection to improve MAI andbeat noise tolerance in OCDMA systems. In: OFC .

87. Benedetto, S. et al . (1994) Coherent and direct-detection polarization modulation system experiment. In:ECOC , Firenze, Italy.


88. Griner, U.N. and Arnon, S. (2004) A novel bipolar wavelength-time coding scheme for optical CDMAsystems. IEEE Photonics Tech. Letters , 16 (1), 332–334.

89. Gu, F. and Wu, J. (2005) Construction of two-dimensional wavelength/time optical orthogonal codesusing difference family. J. Lightw. Technol., 23 (11), 3642–3652.

90. Teixeira, A.L.J. et al . (2001) All-optical time-wavelength code router for optical CDMA networks. In:LEOS, The 14th Annual Meeting of the IEEE .

91. Liang, W. et al . (2008) A new family of 2D variable-weight optical orthogonal codes for OCDMAsystems supporting multiple QoS and analysis of its performance. Photonic Network Communications ,16 (1), 53–60.

92. Liu, M.Y. and Tsao, H.W. (2000) Cochannel interference cancellation via employing a reference correlatorfor synchronous optical CDMA system. J. Microw. & Opt. Tech. Let., 25 (6), 390–392.

93. Yamamoto, F. and Sugie, T. (2000) Reduction of optical beat interference in passive optical networksusing CDMA technique. IEEE Photonics Tech. Letters , 12 (12), 1710–1712.

94. Gamachi, Y. et al . (2000) An optical synchronous M-ary FSK/CDMA system using interference canceller.J. Electro. & Comm. in Japan , 83 (9), 20–32.

95. Yang, C.C. (2007) Optical CDMA passive optical network using prime code with interference elimination.IEEE Photonics Tech. Letters , 19 (7), 516–518.

96. Lin, C.L. and Wu, J. (2000) Channel interference reduction using random Manchester codes for bothsynchronous and asynchronous fiber-optic CDMA systems. J. Lightw. Technol., 18 (1), 26–33.

97. Mustapha, M.M. and Ormondroyd, R.F. (2000) Dual-threshold sequential detection code synchroniza-tion for an optical CDMA network in the presence of multiuser interference. J. Lightw. Technol.,18 (12), 1742–1748.

98. Yang, G.-C. (1994) Performance analysis for synchronization and system on CDMA optical fiber networks.IEICE Trans. on Comm., E77B (10), 1238–1248.

99. Keshavarzian, A. and Salehi, J.A. (2005) Multiple-shift code acquisition of optical orthogonal codes inoptical CDMA systems. IEEE Trans on Comm., 53 (4), 687–697.

100. Sivalingam, K.M. and Subramanian, S. (2005) Emerging optical network technologies . SpringerScience+Business Media Inc, Boston, USA.

101. Goralski, W. (1998) ADSL and DSL Technologies . McGraw-Hill, USA.102. Ohara, K. (2003) Traffic analysis of Ethernet-PON in FTTH trial service. In: OFC .103. Kitayama, K., Wang, X. and Wada, N. (2006) OCDMA over WDM PON – solution path to gigabit

symetric FTTH. J. Lightw. Technol., 24 (4), 1654–1662.

introduction to optical communications copyrighted · pdf fileintroduction to optical ... the...

Documents